Day 4 of AI - slowing down & speeding up
adsbot
Still waiting on the permissions from amazon (LwA). They have contacted me, so it isn’t a dead-end, but the conversation has not been productive yet. They couldn’t figure out which account was mine (despite the email coming from my account). Then when I sent them my credentials, they marked the ticket as resolved without fixing the problem. Typical big company IT issues.
igbot
My auto-post didn’t work yesterday, but it turned out to be because of a scheduling discrepancy between pieces of the bot. Not sure I’ve worked it out. I triggered the post “manually” and it went out to IG/FB/Pinterest. But I’ve also noted things to work on:
Carousel posts don’t work. We need to debug. I’m not even sure what that looks like for Pinterest or FB. Anyway, needs work.
Every “random” post (3 posts generated) from my books has so far been for Mother of Trees. Makes me wonder if there’s a bug.
We can track very little of the post data, and none on Pinterest. I’ve applied for more permissions on Pinterest, and created a little video (that Claude helped me with) to show what I’m trying to do.
Missing Impressions on FB/IG sucks. All I get is the “like” count. I want more.
growthbot
I worked more on this, as it stitches together data from the other bots. That simultaneously improves the other bots while getting this one alive. I think I mentioned that the output goes to email and a slack channel.
Early on, it was giving me recommendations like this:
This Week's #1 Priority
Set up a Kit landing page for Shepherds of Truth with a lead magnet (free chapter or series catch-up guide), then update the Instagram bio link to point to it and add a CTA to every quote graphic posted this week. You have 47 days and no active list-building mechanism — fixing that is more valuable than any content tweak.
It’s a solid recommendation, except that I do have that already. It took many iterations of fumbling to get things even somewhat connected so that the data came through. I know this is a bit much for a blog, but here is the latest iteration of growthbot suggestions:
Summary
With 46 days until Shepherds of Truth launches, there is no active ad campaign running — that's the biggest risk on the board right now. KDP is generating modest but real revenue ($42 in royalties, strong KENP pages in both US and UK), and email open rates on segmented sends are remarkably high (85.7% on 14 recipients), but the main list engagement is soft. Instagram insights are unavailable this week, making optimization nearly impossible — fixing that is urgent.
This Week's #1 Priority
Launch a Facebook ad campaign today — even at $5/day. With 46 days until Shepherds of Truth and no active campaign, you are losing pre-launch momentum every single day. Clone your best-performing past campaign, point it at a Book 1 acquisition or pre-order page, and start building a warm audience you can retarget on April 13. Nothing else on this list matters more right now.
Channel Highlights
• INSTAGRAM: Two posts went out with a combined 13 likes and 3 comments, but reach and impression data are unavailable — quote graphics outperformed trope posts (15.0 vs 4.0 score). Without reach data, it's impossible to assess true performance; diagnosing the insights outage should be immediate.
• PINTEREST: No Pinterest data was included in this snapshot — performance is unknown for the week.
• EMAIL: The full list (1,019 recipients) opened the Beerburg signing email at only 22.6%, well below the account's 52.9% average open rate — subject line and relevance to non-local subscribers likely drove that gap. The 14-recipient 'Market lessons' send hit 85.7% open rate, signaling a highly engaged micro-segment worth nurturing.
• ADS: No active ad campaign is running with 46 days to launch — this is a critical gap with Shepherds of Truth releasing April 13.
• KDP: Strong KENP read-through this week: 12,416 US pages and 7,252 UK pages alongside 32 ebook units and $42.03 in royalties — the UK KU readership is a notable asset worth targeting explicitly.
Optimizer — Improve What's Working
1. Launch a Pre-Order or Launch-Awareness Facebook Ad Campaign Immediately Facebook Ads
There is no active ad campaign with 46 days until Shepherds of Truth releases. Past data referenced a successful FB ads result ('A Strange Win: FB Ads, Profit, and Progress'), meaning there's a proven playbook to activate. Every week without ads is a week of lost pre-launch momentum.
→ Reactivate or clone the best-performing past Facebook conversion campaign, targeting Thaumatropic Roots readers and lookalike audiences. Start with a $5–10/day budget driving to either a pre-order page or a Book 1 free/discounted funnel to build series readership before April 13.
Expected: Begin building a warm audience of series readers who can be retargeted at launch, directly increasing Day 1 sales velocity and KU page reads.
2. Fix Instagram Insights Access Before Next Post Instagram
Reach, impressions, and saves all returned 0 this week — this is almost certainly a data access or API permissions issue, not actual zero performance. Without this data, content optimization is blind.
→ Log into Instagram directly, check Insights on both posts manually, and verify that the analytics tool pulling data has current API permissions. Screenshot actual reach/impression numbers from native IG Insights as a fallback.
Expected: Restore the ability to make data-driven content decisions, especially critical in the 6-week pre-launch window.
3. Segment and Resend a Launch-Teaser Email to Highly Engaged Subscribers Email (Kit)
The 14-recipient 'Market lessons' send hit 85.7% open rate vs. 22–29% on full-list sends. This suggests there's a high-engagement segment (possibly a welcome sequence or a tag for Thaumatropic Roots readers) that responds dramatically better to relevant content.
→ In Kit, create or identify a segment of subscribers tagged as Thaumatropic Roots readers or who have clicked past links related to that series. Send them an exclusive 'first look' email — a chapter excerpt or cover detail from Shepherds of Truth — with a subject line like 'You asked what happens next...'
Expected: Higher open and click rates from your most relevant audience, building anticipation among readers most likely to buy on launch day.
4. Capitalize on UK KENP Readership with Targeted Messaging KDP / Instagram / Email
UK KENP pages (7,252) are nearly 60% of US pages (12,416) — that's a disproportionately large UK readership for an indie author. This audience is actively reading the series right now.
→ Add a UK-targeted interest layer to your next Facebook ad set (target UK Kindle Unlimited subscribers, fantasy readers). Also, add a brief UK callout in your next email — something like 'To my UK readers devouring the series in KU — Book 4 is coming April 13' — to acknowledge and warm that audience.
Expected: Better ad ROI by targeting an already-proven audience geography, and stronger UK launch sales/reads on April 13.
5. Reframe Full-List Email Subject Lines Around Reader Emotion, Not Author Events Email (Kit)
Both full-list sends this week used event-centric subjects ('Beerburg signing tonight,' 'Local Appearance at Library') and landed at 22–26% open rates — well below the 52.9% account average. Non-local subscribers have no reason to open a signing announcement.
→ For the next full-list broadcast, use a subject line tied to reader desire or story emotion rather than author logistics. Example: 'The magic system in Book 4 breaks one of my own rules — here's why' or 'Which Thaumatropic Roots character would you trust with your life?'
Expected: Return open rates closer to the 52.9% benchmark, increasing the number of engaged readers primed for the April 13 launch.
Explorer — New Opportunities
1. Recruit 20 ARC Readers from Your Existing Email List for Shepherds of Truth new_tactic
With 46 days to launch and an email list of ~1,000 subscribers who already know the series, you have a warm pool of potential ARC readers who can post reviews on Amazon and Goodreads on release day. Early reviews are the single biggest launch-day credibility signal for indie fantasy.
First step: Send a dedicated ARC recruitment email this week: 'I need 20 readers to get Shepherds of Truth early — want in?' Link to a simple Kit form or Google Form to collect names. Deliver ARCs via BookFunnel in the last two weeks of March with a clear request to post a review on April 13.
Upside: 15–25 reviews on launch day, dramatically improving Amazon conversion rate and algorithmic visibility.
2. Post a 'Magic System Explainer' Reel or TikTok for BookTok Discovery new_channel
Thaumatropic Roots has an inherently visual, explainable magic system ('Thaumatropic' implies unique mechanics) — exactly the content BookTok and fantasy YouTube Shorts audiences engage with. Found-family and underdog tropes are also BookTok catnip. Short-form video is the highest-reach organic channel available to indie authors right now.
First step: Record a 45–60 second vertical video — no fancy equipment needed — where Steven explains the core magic system rule in Thaumatropic Roots with a hand-drawn diagram or simple visual prop, ending with 'Book 4 drops April 13.' Post to TikTok and as an Instagram Reel simultaneously.
Upside: Even one video with 5,000 views introduces more new readers than a month of feed posts, and magic system content has a high save/share rate in fantasy communities.
3. Pitch a Newsletter Swap with a Complementary Indie Fantasy Author new_tactic
Steven's audience loves found family and underdogs — these are niche but passionate reader identities shared across many indie fantasy communities. A newsletter swap (you feature their book, they feature yours) costs nothing and puts Shepherds of Truth in front of an already-warm fantasy reader audience before launch.
First step: Identify 2–3 indie fantasy authors with email lists of similar size (500–2,000 subscribers) who write found-family or ensemble-cast stories. Reach out via Instagram DM or email this week proposing a swap to run in late March — 3 weeks before launch. StoryOrigin is a free tool to coordinate this formally.
Upside: 100–400 new warm email subscribers or direct clicks to a pre-order/series page with zero ad spend.
4. Create a Goodreads 'Want to Read' Campaign for Shepherds of Truth new_audience
Goodreads is where fantasy readers actively track and discover books — it's a high-intent platform that indie authors underuse. With a release 46 days out, building Want to Read numbers now influences Goodreads' own recommendation algorithm and social proof.
First step: Ensure Shepherds of Truth has a Goodreads listing with cover, description, and April 13 release date. Then ask your email list and Instagram followers in separate posts this week to click 'Want to Read' — frame it as 'Help me show Goodreads this book exists: takes 10 seconds.' Include the direct Goodreads link.
Upside: 50–200 Want to Read adds that improve organic Goodreads discovery and serve as social proof for new readers considering the series.
Anomalies
Instagram reach, impressions, and saves all returned exactly 0 — this is almost certainly a data/API issue, not real performance. Native IG Insights should be checked immediately.
The 14-recipient 'Market lessons' email had an 85.7% open rate — this micro-segment is exceptionally engaged and worth identifying and expanding.
The full-list 'Beerburg signing tonight' email (1,019 recipients) had only a 22.6% open rate, well below the 52.9% account average — event-specific local content is underperforming with the broader list.
UK KENP pages (7,252) are nearly 60% of US KENP pages (12,416) — unusually high UK readership for an indie author that may represent an underutilized geographic marketing opportunity.
Total subscribers in Kit returned as 0 in the snapshot despite recent broadcasts showing 1,019 recipients — this appears to be a data export error and should be verified.
Many of the suggestions are bad because of data not coming in yet. I have FB integration pieces in-progress, but they are also klunky (much like dumping the kdp data). The adsbot is generating data at an improper cadence, and because of that, there is overlapping data in my database, which makes growthbot draw incorrect conclusions. So… things to work out.
But it also had some pretty good suggestions, like poking on my email list to get them to add my book to the “To be read” pile before the book launch. Some of the suggestions don’t make sense to me, but I did tell the agent to think outside the box, so I expect some odd responses.
Next steps
I think I’d like to try a refactoring of my bots. I think I’m going to organize them under a master directory and try loading Claude as a codex in VSCode, and ask it to look at overlaps and refactor. I imagine I will end up with a set of utility functions and slimmer bots. Dunno.
I’ve just about used my credits up for the week, three days in. I’ll look at the pricing for getting more time, but I don’t see my using Claude this much indefinitely, so I don’t think I want to move up to the next package. We will see.
